“Evaluating Multimodal Representations on Visual Semantic Textual Similarity”. New paper at #ecai2020. We present a new language&vision task and dataset for textual inference. #arXiv👉arxiv.org/abs/2004.01894 #dataset 👉 oierldl.github.io/vsts/

This is a pretty neat integration of GPT-3, Wolfram Alpha and Manim to automatically create animations of equations being solved, created by Shreenabh Agrawal

1. Ikerketarako software berri bat garatzea suertatu bazaizu, ziur denboraren zati handi bat eskaini diozula. Zergatik ez berrerabili beste ikerlariek garatutako softwarea? ♻️ Gure lanetarako besteen artikulu zientifikoetan oinarritzen bagara, zergatik ez softwarean? 🤔 #IkerTxio

#ikergazte2021 kongresuan, 'Twitterreko Euskal Komunitatearen Eduki Azterketa Pandemia Garaian' lanagatik joseba.fdl, Iker García-Ferrero, Ander Salaberria eta Jon Ander Campos taldekideek sari berezia irabazi dute! Zorionak!!!

#IkerGazte2021 kongresuan, HiTZ zentroko joseba.fdl, Iker García-Ferrero, Ander Salaberria eta Jon Ander Campos ikergazteek sari berezia irabazi dute! 'Twitterreko Euskal Komunitatearen Eduki Azterketa Pandemia Garaian' lana izan da saritua. Zorionak!!!

IKERGAZTE SARIAK “Twitterreko euskal komunitatearen eduki azterketa pandemia garaian” lanak irabazi du Udalbiltzako saria. Joseba Fernandez de Landaren da lana, baina bere ordez Ander Salaberriak hartu du saria, eta azken honek egin du aurkezpena kongresuan. Iker Gazte

We are hiring! The center has received funding for several ambitious projects, and is offering five PhD positions and six positions for researchers, both pre- and post-doctoral. Please check 👇🏼 hitz.eus/en/recruitment

⚠️Did #ChatGPT cheat on your test? Probably yes Many papers have evaluated ChatGPT on various benchmarks. However, it is important to consider that LLMs might have seen and memorized these datasets during pretraining. Read our latest blog post: hitz-zentroa.github.io/lm-contaminati… 🧵1/5

As a temporal solution, we propose the “LM Contamination Index”: a collaborative effort within the community to identify data contamination in LLMs. We encourage and welcome contributions to the LM Contamination Index. hitz-zentroa.github.io/lm-contaminati… 🧵4/5

We are very happy to announce the publication of our new paper entitled "Grounding spatial relations in text-only language models", with Ander Salaberria and Eneko Agirre @eagirre.bsky.social at HiTZ zentroa (UPV/EHU) . Paper: sciencedirect.com/science/articl… Code: github.com/gazkune/Spatia… 🧵

We present BiVLC, a Bidirectional Vision-Language Compositionality dataset, which allows evaluating image-to-text and text-to-image retrieval simultaneously. Adding text-to-image retrieval substantially changes the conclusions of previous work. arxiv.org/abs/2406.09952

Gaur joseba.fdl-ak "Adimen Artifizialeko metodoak gizarte ikerkuntzarako: analisi demografikoa, jarreren detekzioa eta joera politikoen identifikazioa" izeneko doktore-tesia irakurri du. Zorionak doktore berriari! 🥳🥳

Eta gaur Ander Salaberria-ren txanda izan da. "Integrating Outside Knowledge and Spatial Reasoning in Vision and language Models" izeneko doktore-tesia aurkeztu du. Zorionak doktore berriari! 🥳🥳

WE ARE HIRING!! The HiTZ Center (hitz.eus) at the University of the Basque Country (UPV/EHU) invites applications for several funded research engineering and pre/postdoctoral positions in Natural Language and Speech Processing. hitz.eus/job-offers

I am happy to announce that our latest work on Visually Augmented Language Models has been accepted in #COLING2025 We present Blind-VaLM, a simple approach removing the need for explicit images to visually augment an LM, while still achieving strong performance on VLU and NLU.

Ayer presentamos el proyecto #LUMINOUS en el congreso de la SEPLN 2025, donde tratamos cómo integrar LLMs en entornos de realidad extendida y analizamos los retos presentes en dichas aplicaciones. Eskerrik asko a todes les que se acercaron!